mlsanityData to 14 Oct 2026

DiST-4D: Disentangled Spatiotemporal Diffusion with Metric Depth for 4D Driving Scene Generation

arXiv 2503.15208 ↗First posted 19 Mar 2025cs.CVcited by 40

Not on the map. Why?

Cited by 20 in End-to-end autonomous driving, but placed by its own abstract, which didn’t fit.

Abstract (arXiv)

Current generative models struggle to synthesize dynamic 4D driving scenes that simultaneously support temporal extrapolation and spatial novel view synthesis (NVS) without per-scene optimization. A key challenge lies in finding an efficient and generalizable geometric representation that seamlessly connects temporal and spatial synthesis. To address this, we propose DiST-4D, the first disentangled spatiotemporal diffusion framework for 4D driving scene generation, which leverages metric depth as the core geometric representation. DiST-4D decomposes the problem into two diffusion processes: DiST-T, which predicts future metric depth and multi-view RGB sequences directly from past observations, and DiST-S, which enables spatial NVS by training only on existing viewpoints while enforcing cycle consistency. This cycle consistency mechanism introduces a forward-backward rendering constraint, reducing the generalization gap between observed and unseen viewpoints. Metric depth is essential for both accurate reliable forecasting and accurate spatial NVS, as it provides a view-consistent geometric representation that generalizes well to unseen perspectives. Experiments demonstrate that DiST-4D achieves state-of-the-art performance in both temporal prediction and NVS tasks, while also delivering competitive performance in planning-related evaluations.

Key facts. Cited by 40 papers in the map (data through 14 Oct 2026). Builds on 55 of its 63 references that are in the map. First posted on arXiv 19 Mar 2025.

Builds on

55 of 63 references.

References · 55

  1. 20233D Gaussian Splatting for Real-Time Radiance Field Renderingcited by 6,226
  2. 2022Flow Matching for Generative Modelingcited by 5,702
  3. 2023Adding Conditional Control to Text-to-Image Diffusion Modelscited by 5,503
  4. 2019nuScenes: A multimodal dataset for autonomous drivingcited by 4,523
  5. 2024Scaling Rectified Flow Transformers for High-Resolution Image Synthesiscited by 4,181
  6. 2023Stable Video Diffusion: Scaling Latent Video Diffusion Models to Large Datasetscited by 2,581
  7. 2019Scalability in Perception for Autonomous Driving: Waymo Open Datasetcited by 2,451
  8. 2024CogVideoX: Text-to-Video Diffusion Models with An Expert Transformercited by 2,261
  9. 2023DUSt3R: Geometric 3D Vision Made Easycited by 1,567
  10. 2022Planning-oriented Autonomous Drivingcited by 1,087
  11. 2022Block-NeRF: Scalable Large Scene Neural View Synthesiscited by 724
  12. 2024Open-Sora: Democratizing Efficient Video Production for Allcited by 652
38 more
  1. 2024ViewCrafter: Taming Video Diffusion Models for High-fidelity Novel View Synthesiscited by 403
  2. 2024Vista: A Generalizable Driving World Model with High Fidelity and Versatile Controllabilitycited by 347
  3. 2025GEN3C: 3D-Informed World-Consistent Video Generation with Precise Camera Controlcited by 341
  4. 2022OneFormer: One Transformer to Rule Universal Image Segmentationcited by 328
  5. 2023Driving into the Future: Multiview Visual Forecasting and Planning with World Model for Autonomous Drivingcited by 319
  6. 2023UniSim: A Neural Closed-Loop Sensor Simulatorcited by 297
  7. 2024GeoWizard: Unleashing the Diffusion Priors for 3D Geometry Estimation from a Single Imagecited by 286
  8. 2019Learning Correspondence from the Cycle-Consistency of Timecited by 269
  9. 2023DriveDreamer: Towards Real-world-driven World Models for Autonomous Drivingcited by 261
  10. 2023MagicDrive: Street View Generation with Diverse 3D Geometry Controlcited by 252
  11. 2023OccWorld: Learning a 3D Occupancy World Model for Autonomous Drivingcited by 233
  12. 2024Street Gaussians: Modeling Dynamic Urban Scenes with Gaussian Splattingcited by 221
  13. 2023EmerNeRF: Emergent Spatial-Temporal Scene Decomposition via Self-Supervisioncited by 217
  14. 2024OmniRe: Omni Urban Scene Reconstructioncited by 161
  15. 2023Periodic Vibration Gaussian: Dynamic Urban Scene Reconstruction and Real-time Renderingcited by 143
  16. 2021DriveGAN: Towards a Controllable High-Quality Neural Simulationcited by 125
  17. 2024DriveDreamer4D: World Models Are Effective Data Machines for 4D Driving Scene Representationcited by 123
  18. 2021TransMVSNet: Global Context-aware Multi-view Stereo Network with Transformerscited by 117
  19. 2024Gaussian in the Wild: 3D Gaussian Splatting for Unconstrained Image Collectionscited by 117
  20. 2024UniScene: Unified Occupancy-centric Driving Scene Generationcited by 106
  21. 2023Street-View Image Generation from a Bird's-Eye View Layoutcited by 102
  22. 2023BEVControl: Accurately Controlling Street-view Elements with Multi-perspective Consistency via BEV Sketch Layoutcited by 102
  23. 2024DriveArena: A Closed-loop Generative Simulation Platform for Autonomous Drivingcited by 98
  24. 2023WoVoGen: World Volume-aware Diffusion for Controllable Multi-camera Driving Scene Generationcited by 93
  25. 2016Learning Dense Correspondence via 3D-guided Cycle Consistencycited by 91
  26. 2024OccLLaMA: An Occupancy-Language-Action Generative World Model for Autonomous Drivingcited by 83
  27. 2024OccSora: 4D Occupancy Generation Models as World Simulators for Autonomous Drivingcited by 78
  28. 2022SurroundDepth: Entangling Surrounding Views for Self-Supervised Multi-Camera Depth Estimationcited by 76
  29. 2024MagicDrive3D: Controllable 3D Generation for Any-View Rendering in Street Scenescited by 71
  30. 2024STORM: Spatio-Temporal Reconstruction Model for Large-Scale Outdoor Scenescited by 55
  31. 2024LidarDM: Generative LiDAR Simulation in a Generated Worldcited by 54
  32. 2024DrivingForward: Feed-forward 3D Gaussian Splatting for Driving Scene Reconstruction from Flexible Surround-view Inputcited by 53
  33. 2024FreeVS: Generative View Synthesis on Free Driving Trajectorycited by 53
  34. 2024Feed-Forward Bullet-Time Reconstruction of Dynamic Scenes from Monocular Videoscited by 49
  35. 2024StreetCrafter: Street View Synthesis with Controllable Video Diffusion Modelscited by 49
  36. 2024DreamDrive: Generative 4D Scene Modeling from Street View Imagescited by 35
  37. 2024SubjectDrive: Scaling Generative Data in Autonomous Driving via Subject Controlcited by 32
  38. 2024Hierarchical Temporal Context Learning for Camera-based Semantic Scene Completioncited by 32

50 most cited of 55 All 55 (JSON)

Built on by

Cited by year (2026 to 14 Oct)

Citing papers per year: 2025: 18; 2026: 22.

40 papers.

Sort:

By area:

  1. 20253D and 4D World Modeling: A Surveycited by 67
  2. 2025DriveLaW:Unifying Planning and Video Generation in a Latent Driving Worldcited by 47
  3. 2025OmniNWM: Omniscient Driving Navigation World Modelscited by 33
  4. 2026Latent-WAM: Latent World Action Modeling for End-to-End Autonomous Drivingcited by 33
  5. 2025GaussianDWM: 3D Gaussian Driving World Model for Unified Scene Understanding and Multi-Modal Generationcited by 23
  6. 2025DVGT: Driving Visual Geometry Transformercited by 21
  7. 2025X-Scene: Large-Scale Driving Scene Generation with High Fidelity and Flexible Controllabilitycited by 20
  8. 2025WorldGen: From Text to Traversable and Interactive 3D Worldscited by 19
  9. 2025Challenger: Affordable Adversarial Driving Video Generationcited by 16
  10. 2025WorldSplat: Gaussian-Centric Feed-Forward 4D Scene Generation for Autonomous Drivingcited by 13
  11. 2025Light-X: Generative 4D Video Rendering with Camera and Illumination Controlcited by 12
  12. 2026CoWorld-VLA: Thinking in a Multi-Expert World Model for Autonomous Drivingcited by 12
28 more
  1. 2025CRUISE: Cooperative Reconstruction and Editing in V2X Scenarios using Gaussian Splattingcited by 11
  2. 2025GS-Occ3D: Scaling Vision-only Occupancy Reconstruction with Gaussian Splattingcited by 11
  3. 2025LongDWM: Cross-Granularity Distillation for Building a Long-Term Driving World Modelcited by 9
  4. 20254D Driving Scene Generation With Stereo Forcingcited by 8
  5. 2025Generative AI for Autonomous Driving: Frontiers and Opportunitiescited by 7
  6. 2026Is Your Driving World Model an All-Around Player?cited by 6
  7. 2026RAYNOVA: Scale-Temporal Autoregressive World Modeling in Ray Spacecited by 4
  8. 2026From Articulated Kinematics to Routed Visual Control for Action-Conditioned Surgical Video Generationcited by 4
  9. 2025GeoWorld: Providing Full-frame Geometry Features to Facilitate 3D Scene Generationcited by 3
  10. 2026FAR-Drive: Frame-AutoRegressive Video Generation in Closed-Loop Autonomous Drivingcited by 3
  11. 2026HERMES++: Toward a Unified Driving World Model for 3D Scene Understanding and Generationcited by 3
  12. 2026UniUncer: Unified Dynamic Static Uncertainty for End to End Drivingcited by 2
  13. 2026OmniDrive: An LLM-Choreographed Multi-Agent World Model with Unified Latent Co-Compression for Multi-View Driving Video Generationcited by 2
  14. 2026OpenLongTail: Generative Scaling of Long-Tail Driving Datacited by 2
  15. 2026Diffusion-guided Generalizable Enhancer for Urban Scene Reconstructioncited by 1
  16. 2026Future Dynamic 3D Reconstruction: Toward 3D World Modeling with Disentangled Ego-Motioncited by 1
  17. 2026FrozenDrive: Zero-Shot Text-Guided Driving Scene Generation and Data Augmentation with Parameter-Free Frozen Diffusion Modelcited by 1
  18. 2026Point as Skeleton: Accumulated Point Cloud Enhanced Autoregressive Generation for Closed-Loop Autonomous Driving Simulationcited by 1
  19. 2025FreeGen: Feed-Forward Reconstruction-Generation Co-Training for Free-Viewpoint Driving Scene Synthesiscited by 0
  20. 2025DriveExplorer: Images-Only Decoupled 4D Reconstruction with Progressive Restoration for Driving View Extrapolationcited by 0
  21. 2026VisionNVS: Self-Supervised Inpainting for Novel View Synthesis under the Virtual-Shift Paradigmcited by 0
  22. 2026DriveTok: 3D Driving Scene Tokenization for Unified Multi-View Reconstruction and Understandingcited by 0
  23. 2026DrivingDepth: Sparse-Prompted Pixel-wise Scale Correction for Driving Depth Estimationcited by 0
  24. 2026GaussianDWM++: Language-Grounded 3D Gaussian Driving World Model for Unified Scene Understanding, Editing, and Multi-Modal Generationcited by 0
  25. 2026SPVC: Structured and Panoptic Video Fixing for Cross-Dataset Driving Scene Renderingcited by 0
  26. 2026PRG-Fusion: Orchestrating Generative Priors with Reconstruction Evidence for Driving View Synthesiscited by 0
  27. 2026HelloWorld: Towards Practical Applications of Generative Driving World Modelscited by 0
  28. 2026SymRegFlow: Symmetry-Regularized Flow Matching for Video World Modelscited by 0